Rip a CD to say 256 kbs (which is high!) and you throw away about 75% of the sound-information on the CD.
Play it back, it gets converted to something with as much informationdensity as the orginal lossless CD, so it makes up 4 times the info actually stored.
With clever algorithms this can still sound pleasing, but recoding this again to a different algoritm (again to 256 Kbs) will again throw away 3/4 of the information (which was to a large amount made up by the first decompression).
Because of the shared traits and cleverness of the algoritms (f.i. throw away more info at the frequency extremes, where hearing is lless sensitive) you can not bluntly state that after the second decompression there is only 1/16 of the original info present, far from it, but hopefully you get my drift.
But then again, the proof of the pudding, so how will it actually sound..? Frankly the thoughts mentioned above stopped me from seriously trying it